University of Pretoria

QHW SHAW

PhD THESIS

CHAPTER 2

2.

2.1.

RCC CONSTRUCTION, RCC MIXES, RCC

INSTRUMENTATION & RCC DAMS STUDIED

INTRODUCTION

The research and investigations addressed in this Thesis relate entirely to the
behaviour and performance of Roller Compacted Concrete in dams. In view of the
fact that many of the observations made and the hypotheses presented relate to the
nature of RCC as a material and the methods applied for construction, it is
considered appropriate to provide the reader with some background on RCC in
these areas. Consequently, Chapter 2 presents a background for RCC construction,
describing the typical composition of RCC, isolating specifically the high-paste
RCC that is of particular interest, the methods applied for construction and their
influence on instrumentation readings and the instruments installed.
Each of the five dams whose instrumentation data are investigated during the
course of this study is subsequently introduced.

2.2.

RCC DAM CONSTRUCTION

2.2.1

BACKGROUND

The following description of RCC is partly extracted from Fultons Concrete

Technology. (Ninth Edition). 2009(1), Chapter 24, RCC for Dams, which was written
by the author of this Thesis.
The term Roller Compacted Concrete (RCC) is used to describe a concrete used in
the construction of dams (and pavements), which combines the economical and
rapid placement techniques used for fill dams with the strength and durability of
concrete. As a consequence of the application of high capacity plant and equipment,
it is most suited to use in large-scale construction and for mass concrete works.
Since the early 1980s, RCC has gained general acceptance as an appropriate
material and method for the construction of dams and by the end of 2009, more
than 350 large RCC dams had been completed worldwide.
In the early days, RCC was perceived as a low quality, low strength mass material.
It has since become possible to produce a range of concrete qualities by roller
compaction, with the most common product being a dense, high quality and
relatively high strength concrete.

2.1

University of Pretoria

QHW SHAW

PhD THESIS

In principle, RCC is placed and compacted in 300 mm deep horizontal layers, at

rates often exceeding 3000 m3 per day, allowing construction progress commonly of
around 10 m in height per month. The main benefits of RCC for dam construction
are increased economy and more rapid implementation.
To minimise cementitious materials contents and to take advantage of the fact that
the critical zones of a large dam do not generally experience load for an extended
period after placement, characteristic strengths for RCC are specified at ages of up
to 1 year, and commonly not less than 90 days age.
2.2.2

MODERN RCCS

Modern RCCs are primarily designed in accordance with two different approaches(2):

The overall approach, which relies on the dam body for water-tightness
through high quality concrete and treatment to ensure well-bonded layer and
lift joints.

The separate approach, which relies on an independent impervious barrier,

which is usually placed on the upstream face.

The majority of RCC dams contain mineral admixtures, most commonly fly ash, as
an active constituent of the concrete.
Beyond the basic requirements of strength, a modern RCC mix is defined by the
paste/mortar (p/m) and the sand/aggregate (s/a) ratios, the maximum size
aggregate (MSA) and the modified Vebe time. These parameters essentially relate to
the achievable density (and impermeability), the achievable compaction ratio and
the tendency of the constituent materials to segregate during handling. Under
construction conditions, the aforementioned properties determine workability and
the difference between permeable, stoney RCC, with planes of weakness and a
cohesive, seamless watertight and dense RCC. For the overall approach, mixes
are designed for maximum density, with a paste/mortar ratio of at least 0.37 being
required to achieve a density of 98.5% of the theoretical maximum solid density.
In modern RCC practice, a tendency to use a MSA of 37.5, or 40 mm has developed,
as larger sizes demonstrate a tendency to segregate in an RCC mix during handling
operations. While early RCC testing suggested that a lower sand/aggregate ratio
was optimal for RCC, compared to conventional vibrated concrete (CVC), practical
experience in the interim has demonstrated that quality control and the
maintenance of RCC consistency is much more realistically achieved in an RCC
with a sand/aggregate ratio exceeding 0.35.
The workability of RCC is determined by testing with the Vebe apparatus, which is
modified to include a surcharge mass of 19.1 kg. For workable RCC, the modified
Vebe time should lie between 10 and 20 seconds. For high-workability RCC, a
modified vebe time of 8 to 15 seconds is usually specified. In the case of lean RCC,
the modified Vebe time generally exceeds 30 seconds.

2.2

University of Pretoria

QHW SHAW

PhD THESIS

With a very significant number of approaches attempted during the early years of
development, three primary concepts emerged for the design and construction of
RCC dams:

The lean RCC dam; for which the cementitious materials content is <
100 kg/m3. For such mixes, often only Portland cement is used without
mineral admixtures, or pozzolanic material.

The RCD method (roller compacted dam) unique to Japan; for which the
cementitious materials content is generally 125 kg/m3, but only the hearting
zone of the dam is RCC.

The high-paste RCC dam; for which the cementitious materials content is >
150 kg/m3.

In the case of high-paste RCC, the RCC material itself provides the watertight
barrier and must be designed for an in-situ permeability equivalent to that of
traditional dam mass concrete. The RCC and the associated construction methods
must further be designed to ensure effective bond between layers. Various facing
systems are applied for high-paste RCC dams, but with the simple objective of
creating a good and durable surface finish. Transverse joints are induced at predetermined intervals, which are generally wider than is the case on a conventional
vibrated mass concrete dam.
In early RCC dam construction, a particular problem was recognised as low bond
between successive placement layers. Whilst a relatively high shear friction angle
could generally be assured between layers under all circumstances, low cohesion
and tensile strengths were compounded by high permeability when a new layer was
placed on an excessively mature existing layer. Development in the interim has
included the use of set retarding admixtures and the use of sloped and noncontinuous layer placement methods to ensure the freshness of the underlying RCC
layer when the subsequent layer is placed.
While such practices are only
implemented where required as part of the dam design, the result is a seamless
bond between successive RCC placement layers, with joint properties equivalent to
the parent RCC properties.
2.2.3

RCC MIX COMPOSITION

In the case of lean RCC, the material itself is not designed for impermeability and
consequently the requirements for aggregate gradings are not necessarily as
prescriptive as would be the case for CVC, although density is always important in
the case of a dam.
Lean RCC is also often referred to as a dry consistency mix RCC. With similar water
contents to high-paste RCC, i.e. 100 125 litres/m3, aggregate contents are
obviously relatively high. To ensure consistency and ease of compaction under
construction conditions, lean mix RCC often contains a high proportion of aggregate
fines (often around 8% of the total aggregate content)(3) that form part of the paste
fraction.

2.3

University of Pretoria

QHW SHAW

PhD THESIS

A typical lean RCC comprises the following materials proportions:

Constituents
By Mass
(kg/m3)
By Volume
(litres/m3)

Portland
Cement

Fly Ash

Water

Coarse
Aggregate

Fine
Aggregate

Air

60 - 70

0 - 40

100 - 125

1500

825

21

113

545

300

12

Ignoring the aggregate fines, the typical lean mix above would contain
approximately 140 litres of paste per m3 of concrete. Including 8% fines in the
aggregates, the total paste would be of the order of 200 litres.
In view of the fact that high-paste RCC is designed for impermeability, maximum
density is important and consequently a continuous aggregate grading is applied.
For the latest high-workability RCC mixes, more restrictive aggregate specifications
than required for CVC are applied, with lower compacted void ratios and tighter
restrictions in respect of aggregate shaping and flakiness. For all high-paste RCC,
aggregates of suitable quality for use in a 30 MPa concrete are required.
A typical high-paste RCC comprises the following materials proportions:
Constituents
By Mass
(kg/m3)
By Volume
(litres/m3)

Portland
Cement

Fly Ash

Water

Coarse
Aggregate

Fine
Aggregate

Retarder

60 - 70

140 - 150

100 - 125

1400

800

21

63

113

510

290

While it is common to allow relatively high percentages of sand fines in RCC to

increase the paste volume, ignoring this component, a high paste RCC will comprise
approximately 200 litres of paste and 800 litres of aggregates per m3 of concrete.
2.2.4

RCC CONSTRUCTION

A number of different approaches exist for RCC placement, but the essential
principle is to place and compact 300 mm (compacted) layers as rapidly as
practically possible, creating a monolithic mass either by placing successive layers
before the first set of the previous layer, or by binding layers together with a
bedding mortar, or concrete. RCC is generally placed continuously between
upstream and downstream formwork and the abutments, with no expansion joints
and with induced joints at predetermined intervals to accommodate long-term
shrinkage and creep and thermal contraction due to temperature drop loads.
While it is optimal to place RCC as continuously as possible, without interruptions,
practical circumstances and breakdowns often necessitate breaks, when a cold
joint is formed and the compacted RCC surface must be green-cut and treated
with mortar, or grout before placement above is resumed.

2.4

University of Pretoria

QHW SHAW

PhD THESIS

Plate 2.1: RCC Construction- Dumping, Spreading & Compaction

Plate 2.2: RCC Compaction

Compaction is achieved
with 10 to 15 tonne,
single-drum
vibratory
rollers generally applying 4
passes in either direction
to achieve the target
compaction. The behaviour
of lean RCC and highpaste RCC under vibratory
compaction
are
quite
different; with the former
consolidating to form a
hard and flat surface and
the latter producing a
live surface, particularly
when the set is retarded, into which the

passage of trucks, etc, can make

an impression. While lean, or dry
consistency mix RCC simply
consolidates in the same manner
as a fill under compaction,
consolidation causes paste to be
squeezed through the aggregate
structure and to rise to the
surface
in
high-paste
(and
particularly
high-workability)
Plate 2.3: High Workability RCC
RCC.

2.5

University of Pretoria

2.2.5

QHW SHAW

PhD THESIS

INDUCED JOINTS IN RCC

Joints are induced in RCC at specific cross-sections by de-bonding placement by

between 25 and 100% and thereby creating a localized weakness that will
concentrate cracking consequential to long term temperature drop shrinkage at a
pre-determined location, where it can be isolated with a waterstop. While South
African practice has to date inserted a de-bonding mechanism in every fourth layer
(see Figure 2.1), international practice generally applies de-bonding in every layer.
The early practice of inserting de-bonding systems into the RCC during placement
(see Plate 2.4 and Figure 2.2), but before compaction, has almost universally been
replaced by driving de-bonding systems into compacted RCC (see Figure 2.3 and
Plate 2.5).
UPSTREAM
FACE

DOWNSTREAM
FACE
INSERTED PLATES

CRACK DIRECTORS TO BE
INSTALLED IN EVERY 4th
LAYER ON INDUCED JOINT

Figure 2.1: Induced Joints Cut into RCC every 4th Layer

Plate 2.4: Induced Joints Inserted into RCC with Placement(3)

2.6

University of Pretoria

QHW SHAW

10t SINGLE DRUM

VIBRATORY ROLLER

CRACK DIRECTOR WITH

TIMBER PLANK

PhD THESIS

FRONT-END LOADER

10t SINGLE DRUM

VIBRATORY ROLLER

REMOVE
TIMBER PLANK

UNCOMPACTED RCC

PLASTIC SHEET
WITH PERFORATED PIPE

7t DOUBLE DRUM
VIBRATORY ROLLER
UNCOMPACTED RCC

Figure 2.2: Induced Joint Inserted with RCC Placement(3)

2.7

University of Pretoria

QHW SHAW

PhD THESIS

INSTALLED
CRACK DIRECTORS

ALIGNMENT OF
INDUCED JOINTS

VIBRATING
BLADE

GROOVE FORMED WITH

VIBRATING BLADE FORCED
INTO FRESHLY COMPACTED
RCC

BLADE WITHDRAWN LEAVING

SHEET IN GROOVE.
GROOVE CLOSED WITH
UPSTREAM-DOWNSTREAM
ROLLER COMPACTION

FRESHLY
COMPACTED
RCC

GALVANISED
STEEL SHEET

250

GALVANISED
STEEL SHEET

Figure 2.3: Induced Joint Inserted into Compacted RCC

2.8

University of Pretoria

QHW SHAW

PhD THESIS

Plate 2.5: Induced Joint Inserted into Compacted RCC

Although several methods and systems are used for de-bonding RCC to create an
induced joint, all result in an increased compressibility on the induced joint,
compared to the adjacent RCC. While the construction process causes a
disturbance of the RCC on either side of the induced joint, the insertion of folded
plastic sheeting, geotextile material, or a folded galvanized steel sheet, implies that
the aggregate-to-aggregate contact within the RCC is broken. As the presence of the
de-bonding system implies that the RCC structure cannot be as effectively
redeveloped during the subsequent re-compaction, the increased compressibility is
undoubtedly not only the consequence of a compressible joint filler, but also a
slightly more open structure within the concrete on either side.
A consequence of the evident compressibility of the de-bonded areas of the induced
joints is local exaggerated movement/closure during thermal expansion of the RCC.
As the temperature of the RCC rises with the evolution of hydration heat, it will
experience expansion, which will be restrained by the continuity of the placement in
a direction parallel to the dam axis (left to right bank direction). Due to the fact that
2.9

University of Pretoria

QHW SHAW

PhD THESIS

joint inducers are only installed in every second, or fourth layer in the dams
investigated as part of this study, their presence will not reduce the overall
resistance of the RCC mass to thermal expansion, nor will it cause any perceptible
increase in the overall compressibility of the full joint, but it will undoubtedly give
rise to increased local compression across the actual de-bonded section of the joint

2.3.

RCC INSTRUMENTATION

2.3.1

GENERAL

The instrumentation installed in the dams addressed in this study was designed to
fulfil two specific purposes; to monitor the overall performance of the dam structure
with a view to ensuring continued dam safety and to monitor the behaviour of the
constituent RCC.
The dam safety instrumentation comprised pendulums,
displacement survey systems, load cells, pore pressure meters and seepage
measurement weirs, while the RCC-specific instrumentation comprised long-basestrain-gauge-temperature-meters (LBSGTMs), strain gauges, temperature gauges
and thermocouples.
Instrumentation installed in RCC was generally developed for use in soils and for
monitoring geotechnical structures, etc. Accordingly, it is of a robust construction
and only instruments and measurement systems that have proved accurate and
reliable in conditions in which heavy equipment operates are installed in RCC
dams.
2.3.1

THE INSTRUMENTS

The Long-Base-Strain-Gauge-Temperature-Meters (LBSGTMs) and the strain

gauges installed at all of the dams addressed in this study were vibrating-wire type
gauges, a system that is acknowledged for its long-term stability and accuracy(5).
A long-base-strain-gauge-temperature-meter is essentially a long strain, or
deformation gauge that is installed across (perpendicular to) the alignment of an
induced joint. These instruments are correspondingly aligned parallel to the axis of
the dam structure. The LBSGTM was developed by Geokon for use in RCC dam
construction on the request of the South African Department of Water Affairs
(DWA). The gauge is a vibrating wire meter that measures deformation across a
distance of between 600 mm and 1 m. The long-base was considered necessary to
ensure that the crack on the induced joint passes between the flanged end plates of
the gauge. Initially, a gauge length of 1 m was considered necessary, but it has
since become evident that 600 700 mm is quite adequate. The Geokon Model
4430 gauge is specified for a measurement sensitivity of 0.01 mm and an accuracy
of 0.05 mm. The gauge measures temperature, as well as deformation, and the
indicated deformations must be adjusted for temperature.

2.10

University of Pretoria

QHW SHAW

PhD THESIS

Plate 2.6: Long-Base-Strain-Gauge- Temperature-Meter (LGSGTM)

The first gauges used by DWA were tested under the direct action of a 10 tonne
vibratory roller(6) and were demonstrated to remain operational and to continue
providing accurate measurement no matter how roughly they were treated. The
continued operation of the vast majority of these gauges at Wolwedans Dam after
some 20 years is testament to their robustness.
The Geokon Model 4210 gauge used for measuring strain is a particularly robust,
all-stainless steel-cased instrument that is designed for direct installation into
concrete with large aggregate and indicates a measurement sensitivity of
0.4 microstrain and an accuracy of 1%. The gauge also measures temperature
and the gauge readings must be adjusted to indicate actual concrete strain.

Plate 2.7: Strain Gauge

2.11

University of Pretoria

2.3.2

QHW SHAW

PhD THESIS

INSTRUMENT INSTALLATION

For simplicity and to ensure minimum interruption to RCC placement,

instrumentation to monitor RCC
behaviour is generally installed at
a number of specific elevations
over the height of an RCC dam,
ideally when planned stoppages
will anyway occur for galleries, etc.
Stoppages generally correspond
with cold lift joints, where the
RCC is allowed to set fully. A
number of methods have been
used to create trenches and slots
in the RCC surface for the
instrumentation and associated
cabling and this has sometimes
been done by excavation after Plate 2.8: LBSGTM in Slot
completion of the layer and other
times by cutting slots with a vibrating blade during the process of completing the
layer surface. In either case, the cabling is laid out, the instruments are installed
and the trenches and slots are
backfilled with a structural concrete.
Resumption of RCC placement above
can often be delayed for a further
number of days, but critical data
collection starts as soon as the
respective gauge is covered with a
layer of RCC.
LBSGTMs
are
installed
in,
or
immediately above, the layer in which
the induced joint de-bonding medium
is installed, to ensure that the crack
occurs relatively centrally on the
Plate 2.9: Instrumentation Cable Trench
gauge.
As a consequence of locating the LGSTMs immediately above an induced joint debonding mechanism, the initial deformation readings during the build up in
temperature due to hydration reflect the compressibility of the joint inducer system
installed. While this will result in a distortion and will not provide a true
representation of the actual strain within the continuous RCC in compression
above and below the de-bonding system while the temperatures are elevated, the
related impact will dissipate once the joint experiences any tension and/or
cracking.

2.12

University of Pretoria

2.4.

QHW SHAW

PhD THESIS

THE RCC DAMS STUDIED

The central focus of the work presented in this Thesis is the data recovered from
instrumentation installed in the Wolwedans and Knellpoort Dams in South Africa,
ine Dam in Turkey and Wadi Dayqah Dam in Oman. On the basis of the
interpretation of this data, the behaviour of the constituent RCC is observed over
periods varying from several months to several years. Behavioural observations
from Changuinola 1 Dam, currently under construction in Panama, are also
addressed herein.

2.5.

WOLWEDANS DAM

2.5.1.

INTRODUCTION

Wolwedans Dam was completed in early 1990 and was the first RCC dam in the
world to rely fully on three-dimensional arch action for stability. The dam is 70 m
high, has a crest length of 270 m, an upstream face arch radius of 135 m and
comprises approximately 200 000 m3 of concrete. The dam was constructed with
induced joints at 10 m spacings, de-bonding every 4th layer with inducers installed
during placement and the RCC was placed in October and November of 1988 and
between May and November of 1989. The first season involved the placement of the
bottom 15 m of the dam, while the second encompassed the placement of the
remaining 55 m. RCC placement for Wolwedans was achieved over a total period of
approximately 7.5 months, with a peak daily rate of 2994 m3 (7).
The dam first filled to capacity during 1992 and the induced joints were grouted
during late winter in 1993. With a full storage volume of 64% of the Mean Annual
Runnoff (MAR), a high supply assurance requirement and a relatively high
catchment rainfall, Wolwedans Dam is subject to consistently high water levels.
The dam has spilt on several occasions to date and discharged a relatively large
flood just two years after completion.
The composition of the high-paste Wolwedans RCC mix was as follows(7):
Constituents
By Mass
(kg/m3)
By Volume
(litres/m3)

Portland
Cement

Fly Ash

Water

Coarse
Aggregate

Fine
Aggregate

Air

58

136

100

1510

625

18.5

63.5

100

565

240

13

Ignoring aggregate fines, the Wolwedans RCC comprised 182 litres of paste and 805
litres of aggregates. The blend of 75% crusher and 25% pit sand contained only
1.3% of fines and accordingly, the paste volume was only increased by 3 litres/m3.
With a low sand/aggregate ratio of just 0.3, the paste/mortar ratio was
correspondingly high, at almost 0.44.

2.13

University of Pretoria

QHW SHAW

A basic layout of Wolwedans Dam is provided in Figure 2.4.

2.14

PhD THESIS

Figure 2.4:
Layout of Wolwedans Dam(7)

University of Pretoria

2.5.2.

QHW SHAW

PhD THESIS

WOLWEDANS INSTRUMENTATION

Being one of the first two structures of its type and dependent on arching for
stability, a very comprehensive network of structural and thermal monitoring
instrumentation was installed within the body of Wolwedans Dam (see Figure 2.11
at the end of this Chapter and Figures C1 to C4 in Appendix C). Virtually all of
the instrumentation installed during construction is still functional and a central,
remotely interrogated control system allows real-time monitoring of the dams
performance and behaviour.
LBSGTMs were installed across all induced joints at four different levels and these
instruments formed the core of the system for monitoring the thermal/structural
behaviour of the dam structure. With a lowest foundation elevation of RL 33 m and
a NOC elevation of RL 103 m, instrumentation was installed at elevations
RL 40.25 m, RL 52.25 m, RL 66.25 m and RL 84.25 m.
The lowest level of instrumentation was installed approximately at mid-height of the
first RCC cast during October and November 1988.
The second level of
instrumentation was installed just 4.25 m above the bottom level of the RCC placed
during 1989. The third level of instrumentation was installed at approximately mid
height, 33 m above lowest foundation level. The top level of instrumentation was
installed at the level of the top gallery, implying that instruments were installed 2 m
from an external surface on one side and 2.75 m from the gallery on the other.

Plate 2.10: Wolwedans Dam

2.5.3.

IMPORTANT INFLUENCES ON RECORDED BEHAVIOUR

In view of the fact that the dam structure at Wolwedans was constructed in two
distinct parts(7), separated by a substantial break, it is not considered that the full
structure behaves entirely monolithically with respect to temperature and hydration
2.15

University of Pretoria

QHW SHAW

PhD THESIS

heat dissipation. Two specific issues are considered of importance in this regard;
construction of the first part during the summer months and initiation of the latter
part in winter and the fact that the instrumentation installed at elevation
RL 40.25 m suggests that almost all of the hydration heat from the bottom section
had been dissipated by the time that the construction of the upper section was
initiated. Correspondingly, the instrumentation data at elevations RL 40.25 m and
RL 52.25 m are likely to have been influenced significantly by the extended
interruption in construction that occurred at RL 48 m.
With only 2 rows of LBSGTMs and surface cover of just 2 m to the outside and
2.75 m to the gallery, it is considered that the readings in the highest level of
instrumentation (RL 84.25 m) will substantially reflect surface zone temperatures
and effects. In the third level of instrumentation (RL 66.25 m), however, foundation
restraint will not be a significant influence, while the dam wall thickness is
approximately 21 m, which the instrumentation records demonstrated to be
sufficient to limit the core temperature variation over a typical annual cycle to
approximately 2oC, while still allowing all of the hydration heat to be dissipated
within 2 years after dam completion.
The induced joints at Wolwedans Dam were grouted in two phases, between July
and November 1993. With the impounded water level dropped by 8 m during the
latter part of the period, grouting to mid dam height (RL 66.25 m) was completed
over the winter months of June to August. With the dam filled to capacity once
more, the top half of the structure was subsequently grouted over the Spring and
early Summer months of September to November. It is significant to note that while
a net upstream crest movement of the order of 2.5 mm was recorded when the
water level was drawn down, equivalent displacements in the upper gallery were of
the order of only 0.5 mm and no associated movement was really apparent on any
of the induced joint instruments. At the time that the induced joints were grouted,
it is apparent from the displacement data records that the dam crest was displaced
downstream by a maximum of well over 10 mm, indicating that the structure was
already subject to a significant temperature drop and that the grouting merely filled
open joints and did not serve a significant purpose in alleviating the impact of longterm temperature-related loading.

2.6.

KNELLPOORT DAM

2.6.1.

INTRODUCTION

Knellpoort Dam is also in South Africa and was designed in parallel with
Wolwedans, but construction was initiated earlier and completed over a shorter
period, as concrete volumes were relatively small. Although this 50 m high dam is
defined as an arch/gravity structure, arching is only incurred under extreme
loading conditions and, as an off-channel storage dam, with a large capacity
compared to the natural catchment inflow, hydrostatic loadings are generally low.

2.16

University of Pretoria

QHW SHAW

PhD THESIS

The main part of the dam is aligned on a circular arch and has a vertical upstream
face and a downstream face sloped at 0.6 horizontal to 1 vertical. The left flank
comprises a straight conventional gravity structure. The combined crest length
measures 200 m.
The 60 000 m3 of concrete comprising Knellpoort Dam was placed over a single
winter season during 1988. Induced joints were aligned radially at a 10 m spacing
on the upstream face, de-bonding every 4th layer with inducers installed during
RCC placement.

Plate 2.11: Knellpoort Dam(8)

The composition of the RCC mix placed at Knellpoort Dam was as follows(8):
Constituents
By Mass
(kg/m3)
By Volume
(litres/m3)

Portland
Cement

Fly Ash

Water

Coarse
Aggregate

Fine
Aggregate

Air

61

142

108

1610

685

19.5

66

108

555

245

6.5

Ignoring aggregate fines, the Knellpoort RCC comprised approximately 195 litres of
paste and 800 litres of aggregates. Again, a low sand/aggregate ratio of just 0.3
gives rise to a high paste/mortar ratio of 0.44.
A basic layout of Knellpoort Dam is provided in Figure 2.5.

2.17

University of Pretoria

QHW SHAW

2.18

PhD THESIS

Figure 2.5:
Layout of Knellpoort
Dam(8)

University of Pretoria

QHW SHAW

PhD THESIS

2.6.2. KNELLPOORT INSTRUMENTATION

The instrumentation installed at Knellpoort Dam was essentially identical to that
installed at Wolwedans. Despite the
lower dam height, the reduced
dependence on arching and the
inclusion of only a single gallery,
instrumentation was installed on four
separate levels.
However, unlike
Wolwedans Dam, Knellpoort was
constructed
largely
during
a
particularly cold winter, with built-in
temperatures frequently below 15C.
Plate 2.12: Knellpoort Dam(9)
2.6.3.

IMPORTANT INFLUENCES ON RECORDED BEHAVIOUR

Knellpoort Dam was constructed primarily over a particularly cold winter in 1988.
The temperature at which the RCC at any particular location in the dam body was
placed, or effectively insulated by RCC placed above, was frequently below 15C. In
view of the fact that the long-term temperature at the core of the dam structure
varies seasonally between 13 and 16C, the difference between the average longterm core temperature and the average built-in temperature is minimal, as
indicated in Figure 2.6.

35

RCC TEMPERATURE ( C)

PEAK HYDRATION TEMPERATURE

30

25
"CORE" TEMPERATURE
LONG-TERM MINIMUM TEMPERATURE

20

15
PLACEMENT, OR "BUILT-IN" TEMPERATURE

10
1988

1989

1991

1990

1992

1993

1994

DATE

Figure 2.6: Typical RCC Temperature History for Knellpoort Dam(8)

2.19

University of Pretoria

2.7.

INE DAM

2.7.1.

INTRODUCTION

QHW SHAW

PhD THESIS

ine Dam is currently under

construction in south-western
Turkey. Placement of RCC is
restricted
to
the
winter
months of the year and
limited each year on the basis
of
available
State
fund
allocations. By April 2009,
almost 1.4 million m3 of the
required
total
of
1.65 million m3 of RCC had
been placed. The dam is a
gravity structure with a
Plate 2.13: ine Dam
maximum height of 136.5 m
and a crest length of approximately 300 m.
While the topography could
accommodate an arch structure, the foundation rockmass conditions were not
considered suitable. As a consequence of the dams location in an area of relatively
high seismic risk, the structure has a wide base and the RCC strength
characteristics are determined by the related requirements under seismic loading.
Induced joints at ine Dam were installed in the compacted RCC at intervals of
24 m along the length of the main body of the wall, de-bonding every 4th layer. On
the extremes of the flanks, where the dam height reduces, a closer joint spacing
was applied. The RCC for ine Dam is zoned, with the an upstream impermeable
zone containing 85 kg/m3 cement and 105 kg/m3 fly ash and the remaining bulk of
the wall structure containing 75 kg/m3 cement and 95 kg/m3 fly ash.
The compositions of the two RCC mixes placed at ine Dam were as follows(10):
Constituents

Portland
Cement

Fly Ash

Water

Coarse
Aggregate

Fine
Aggregate

Air

By Mass
(kg/m3)

85

105

115

1406

791

By Volume
(litres/m3)

27

46

115

516

290

Fines
(l/m3)

Aggregate
(l/m3)

Paste/
Mortar

Sand/
Aggregate

49

806

0.40

0.36

D10 Mix

Net Paste
(l/m3)
193

2.20

University of Pretoria

Constituents

QHW SHAW

PhD THESIS

Portland
Cement

Fly Ash

Water

Coarse
Aggregate

Fine
Aggregate

Air

By Mass
(kg/m3)

75

95

120

1590

586

By Volume
(litres/m3)

24

41

120

586

218

11

Fines
(l/m3)

Aggregate
(l/m3)

Paste/
Mortar

Sand/
Aggregate

36

804

0.46

0.27

D05 Mix

Net Paste
(l/m3)
185

A basic plan layout of ine Dam is provided in Figure 2.7.

2.7.2.

INE INSTRUMENTATION

A comprehensive network of monitoring instrumentation was installed in ine Dam,

including 200 LBSGTMs, 50 separate temperature gauges, 3 pendulum lines, 30
strain gauges, 30 surface crack meters, 13 piezometers and 6 seepage
measurement sites. The dam contains 3 separate, horizontal galleries at elevations
147.5 mASL, 185 mASL and 232 mASL and instrumentation was installed at each
of these levels, as well as at elevation 210 mASL.
For the purposes of evaluating the thermal performance of the RCC at ine Dam,
data from the LBSGTMs, the strain gauges and the various temperature meters
were analysed. The LBSGTMs were ascribed an SGT designation, while the strain
gauges were ascribed an SGA designation. Figure 2.12 at the end of this Chapter
and Figures C5 to C7 illustrate the basic layout of the installed instrumentation on
a typical induced joint. The SGA strain gauges installed at ine Dam were aligned
in an array in an upstream-downstream direction to measure strain perpendicular
to the dam axis.
2.7.3.

IMPORTANT INFLUENCES ON RECORDED BEHAVIOUR

As previously mentioned, the RCC for ine Dam was placed between October and
April of each year from 2004. Each season saw the placement of approximately
300 000 m3 of RCC, with the levels listed in Table 2.1 being achieved at the end of
each placement in April. The surface of the RCC at this level was subsequently
exposed to the elements, until placement resumed the following October.

2.21

QHW SHAW

Figure 2.7:
Illustrative Layout of
ine Dam(11)

University of Pretoria
PhD THESIS

2.22

University of Pretoria

QHW SHAW

The dam structure, constructed in

programme, was consequently the
undertaken to establish whether any
might be developed and to establish
insulation may be required.

PhD THESIS

accordance with the actual construction

subject of a detailed thermal analysis,
consequential deleterious thermal stresses
whether any RCC cooling and/or thermal

This thermal analysis will not be addressed in detail in this study, as the related
results have no specific relevance. However, it is important to note that a very good
correlation was evident between the temperatures within the dam predicted by the
finite element thermal model and those measured on the prototype structure and
this is illustrated in Chapter 4.
Table 2.1: RCC Progress at ine Dam
Placement Period

Level Achieved
(m ASL)

October 2004 April 2005

147.50

October 2005 April 2006

166

October 2006 April 2007

187

October 2007 April 2008

208.25

October 2008 April 2009

232

October 2009 October 2010

February 2005

265 (completed)

April 2006

Figure 2.8: ine Dam Construction Progress

2.23

University of Pretoria

QHW SHAW

PhD THESIS

April 2007

April 2008

February 2010
Figure 2.8: ine Dam Construction Progress continued

2.8.

WADI DAYQAH DAM

2.8.1.

INTRODUCTION

Construction of Wadi Dayqah Dam in Oman was recently completed (2009). The
dam comprises an 80 m high RCC gravity structure and a 40 m high rockfill saddle
embankment. With a crest length of a little under 400 m, the RCC structure
comprises a total concrete quantity of approximately 650 000 m3 and the RCC was
essentially placed between February and December 2008. The dam is aligned on a
curve, with a radius of 500 m to the upstream face and induced joints are arranged

2.24

University of Pretoria

QHW SHAW

PhD THESIS

radially, generally at a spacing of 15 m measured on the upstream face. Induced

joints were installed in every 2nd layer of RCC after compaction.

Plate 2.14: Wadi Dayqah Dam

The RCC for Wadi Dayqah Dam was zoned, with an upstream impermeable
15 MPa RCC zone containing 126 kg/m3 cement and 54 kg/m3 ground Limestone
and the remaining bulk of the wall structure (12 MPa RCC) containing 112 kg/m3
cement and 48 kg/m3 ground Limestone.
The compositions of the two RCC mixes placed at Wadi Dayqah were as follows(12):
Constituents

Portland
Ground
Cement Limestone

Water*

Coarse
Aggregate

Fine
Aggregate

Retarder

Zone 1 - 15 MPa Mix

By Mass
(kg/m3)
By Volume
(litres/m3)
Net Paste
(l/m3)
166.5

126

54

137

1200

944

1.76

41

21

103

455

378

1.5

Fines
(l/m3)

Aggregate
(l/m3)

Paste/
Mortar**

Sand/
Aggregate

51

833

0.31

0.45

*-

the free water content was 103 litres/m3, but an additional quantity of 34 litres was
required due to aggregate absorption.

** -

including aggregate fines increases the paste to 217.5 litres/m3 and the p/m to 0.40.

2.25

University of Pretoria

Constituents

QHW SHAW

Portland
Ground
Cement Limestone

PhD THESIS

Water*

Coarse
Aggregate

Fine
Aggregate

Retarder

Zone 2 12 MPa Mix

By Mass
(kg/m3)
By Volume
(litres/m3)
Net Paste
(l/m3)
153.5

112

48

131

1227

960

1.76

37

19

96

463

384

1.5

Fines
(l/m3)

Aggregate
(l/m3)

Paste/
Mortar**

Sand/
Aggregate

52

847

0.29

0.45

*-

the free water content was 96 litres/m3, but an additional quantity of 35 litres was
required due to aggregate absorption.

** -

including aggregate fines increases the paste to 205.5 litres/m3 and the p/m to 0.38.

2.8.2.

WADI DAYQAH INSTRUMENTATION

The instrumentation installed in Wadi Dayqah RCC gravity dam in Oman was less
comprehensive than was the case for the previous examples. With the same types
of instruments, only a single LBSGTM was installed across each of the induced
joints in the centre of the section at two elevations, namely 135 mASL and
150 mASL. Five levels of concrete temperature meters and external temperature
gauges were also installed on three specific cross sections; one in the spillway
section and one on the non-overspill section on either flank. The typical layouts of
the temperature meters and the LBSGTMs are illustrated on Figures 2.13 and 2.14
at the end of this Chapter.
2.8.3.

IMPORTANT INFLUENCES ON RECORDED BEHAVIOUR

Placement of the RCC for Wadi Dayqah Dam took place between February 2008 and
July 2009 in temperatures varying between 25 and 45C.
However, as a
consequence of the dam having a spillway with Roberts crest splitters, a
substantial slowing in the pace of construction occurred once the base of the
spillway crest was reached in November 2008. The vast majority of the RCC
comprising the dam was accordingly placed in just 10 months.
With a better picture of the behaviour expectations of the RCC and a relatively
small budget for instrumentation, it was considered most appropriate to install a
single LBSGTM across all joints in the centre of the dam section at two levels. With
this configuration, it was understood that only the behaviour of the core zone RCC
would be monitored.

2.26

University of Pretoria
QHW SHAW

2.27

PhD THESIS

Figure 2.9:
Illustrative Layout of
Wadi Dayqah Dam(13)

University of Pretoria

QHW SHAW

PhD THESIS

Whereas all of the other dams for which instrumentation data were reviewed
contain high-paste RCC for which a relatively high proportion of fly ash was used,
the RCC of Wadi Dayqah comprised a relatively low cement content (112 kg/m3) in
combination with a ground limestone filler (48 kg/m3) and a 0.44 sand/aggregate
ratio. Furthermore, 34% of the sand fraction comprised crushed limestone. As a
consequence, the final RCC contained over 13% fines.
Three further factors are considered of importance in respect of the measured
behaviour of the RCC at Wadi Dayqah Dam and these are the rounded particle
shape of the natural gravel coarse aggregate, the high water absorption
characteristics of the aggregates (approximately 35 litres/m3) and the fact that the
RCC temperature was artificially cooled by approximately 15C before placement.

Plate 2.15: Wadi Dayqah Dam January 2009

2.9.

CHANGUINOLA 1 DAM

2.9.1.

INTRODUCTION

Placement of the RCC for Changuinola 1 Dam in Panama commenced in December

2009, with completion scheduled for February/March 2011. The dam is a 105 m
arch/gravity structure comprising approximately 890 000 m3 of RCC. The upstream
face arch radius is 525 m, the induced joints are spaced at 20 m intervals, with
2.28

University of Pretoria

QHW SHAW

PhD THESIS

inducers installed in every 2nd layer after compaction and the downstream face
slope varies from 0.5 H to 1 V in the centre to 0.7 H to 1 V on the flanks.
For Changuinola 1, a high-workability RCC is being applied, with a first set
retarded typically to 20 hours. The mix strength requirements are determined
primarily by a target direct vertical tensile strength of 1.2 MPa and peak RCC
placement rates should exceed 120 000 m3 per month.
The composition of the high-workability RCC mix applied for Changuinola 1 is
follows:
Constituents

Portland
Cement

Fly Ash

Water

Coarse
Aggregate

Fine
Aggregate

Retarder

By Mass
(kg/m3)

70

145

119

1282

888

3.44

By Volume
(litres/m3)

22

60

119

462

334

Fines
(l/m3)

Aggregate
(l/m3)

Paste/
Mortar

Sand/
Aggregate

35

799

0.375

0.42

Net Paste
(l/m3)
201

Including the aggregate fines within the paste increases the p/m to 0.44.
2.9.2.

CHANGUINOLA 1 INSTRUMENTATION

The instrumentation to be installed in Changuinola 1 Dam is very similar to the

arrangements described above. At the time of writing, however, only one level of
instruments had been installed and the while these were indicating the same
patterns as for Wolwedans and Knellpoort, the record was of not of adequate length
to be used for the purposes of the work addressed herein.

Figure 2.10: Changuinola FE Model Illustrating Final Layout

2.29

University of Pretoria

2.9.3.

QHW SHAW

PhD THESIS

IMPORTANT INFLUENCES ON RECORDED BEHAVIOUR

With almost 3 m of rainfall per annum on site and a wet month containing 25 days
of rain and a dry month 16 days, the weather conditions on site represent a very
significant factor in relation to the construction of Changuinola 1 Dam. While RCC
dam construction offered very significant advantages in respect of the river
diversion, compared to other dam types, the final programme and the shear size of
the river implied that the structure was constructed to approximately 40 m height
in two sections, with a formed joint in between (see Plate 2.16). Another factor of
specific relevance in the case of Changuinola 1 Dam is the temperate climatic
conditions, with average monthly temperatures varying only by 4C year-round.

Plate 2.16: Construction Progress at Changuinola 1 Dam on 25th June 2010

2.10.

INSTRUMENTATION LAYOUTS

2.10.1.

WOLWEDANS DAM

Figure 2.11 illustrates the typical layout of the instrumentation installed in

Wolwedans Dam. Additional plan layouts of each of the instrumentation levels are
provided for illustration on Appendix C.

2.30

University of Pretoria

RL 98.0 m

QHW SHAW

PhD THESIS

4000
200 MM DIA PENDULUM
SHAFT

FOURTH LEVEL RL 84.25 m

OF INSTRUMENTS

Long-Base-Strain-GaugeTemperature Meters

RECESS FOR
PENDULUM

RL 66.25 m
THIRD LEVEL
OF INSTRUMENTS

SECOND LEVEL RL 52.25 m

OF INSTRUMENTS

RECESS FOR
PENDULUM

MAIN DRAINAGE
GALLERY

FIRST LEVEL
RL 40.25 m
OF INSTRUMENTS

PIEZOMETER &
TEMPERATURE
GAUGES
200 MM DIA
PENDULUM
SHAFT

Figure 2.11: Illustrative Section Indicating Instrumentation(7)

(from 1987/88 Hand Drawn Plan)

2.31

University of Pretoria

2.10.2.

QHW SHAW

PhD THESIS

INE DAM

Figure 2.12 illustrates the typical instrumentation installed in ine Dam. The
layouts for the instrumentation at El 147.5 mASL, El 184.25 mASL & El 208.5
mASL are included in Appendix C.

El 232.00 mASL

El 208.25 mASL

El 185.25 mASL

El 147.50 mASL

FIGURE 2.12: Typical Instrumentation Installed in ine Dam(11)

2.32

University of Pretoria

QHW SHAW

PhD THESIS

2.10.3. WADI DAYQAH DAM

Figures 2.12 and 2.14 illustrate the typical instrumentation installed in Wadi
Dayqah Dam.

Figure 2.13: Typical Instrumentation - Spillway Section Wadi Dayqah Dam(13)

2.33

University of Pretoria

QHW SHAW

PhD THESIS

Figure 2.14: Typical Instrumentation - NOC Section Wadi Dayqah Dam(13)

2.34

University of Pretoria

2.11.

QHW SHAW

PhD THESIS

REFERENCES

[1]

Owens, G. Fultons Concrete Technology. Chapter 8. Ninth Edition.

Cement & Concrete Institute. Midrand. RSA. 2009.

[2]

ICOLD. Committee on Concrete for Dams. Roller Compacted Concrete

Dams. State of the Art and Case Histories. ICOLD Bulletin 126. 2003.

[3]

Schrader, EK. Roller Compacted Concrete. Chapter 20. Concrete

Construction Engineering Handbook. Second Edition. Edited by Nawy,
EG. CRC Press. New Jersey. 2008.

[4]

Shaw, QHW. DWS 740. Standard Specification. Roller Compacted

Concrete for Dams. Department of Water Affairs & Forestry. Second
Edition. August 2005.

[5]

McRae, JB & Simmonds, T. Long-Term Stability of Vibrating Wire

Instruments: One Manufacturers Perspective. Proceedings of 3rd
International Symposium on Field Measurements in Geotechnics. Field
Measurements in Geomechanics. Oslo. September 1991.

[6]

Shaw QHW, & Maartens, W.W. Department of Water Affairs Internal

Report. Construction and Grouting of the Wolwedans Dam Test Section.
Un-numbered DWAF Report No. Mossel Bay. October 1988.

[7]

Shaw QHW, Geringer JJ & Hollingworth F. Department of Water

Affairs Internal Report. Wolwedans Dam Completion Report. DWAF
Report No. K200/02/DE01. Pretoria. April 1993.

[8]

The Department of Water Affairs & Forestry Internal Report. Knellpoort

Dam Completion Report. Information Report No. D203/39/DD04.
DWAF, Pretoria. 1991.

[9]

SANCOLD. Large Dams and Water Systems in South Africa. SANCOLD.

CTP-Book Printers. Cape Town, South Africa. 1994.

[10]

zkar Construction Internal Report. Work Reports. Quality Control Unit.

Ankara, Turkey. July 2005 February 2008.

[11]

Geoconsult. Gibb. ARQ. ine RCC Dam. Phase 2 Design Report. Vol.4 of
4. Drawings. zkar Construction. Ankara, Turkey. January 2000.

[12]

Vinci & CCC Construction JV. Wadi Dayqah Dam. Quality Control
Records. Quriyat, Oman. February 2008.

[13]

Wadi Dayqah Dam JV. Wadi Dayqah Dam. Drawings for Construction.
Sultanate of Oman. M.R.M.E.W.R. Muscat, Oman. August 2006.

2.35